DailyDirt: Games Played By Computers

from the urls-we-dig-up dept

Artificial intelligence projects have come a long way with algorithms that can beat some of the best humans at poker and chess and a whole bunch of other games. There are still a few games that people can clearly play better than computers, but that list is getting shorter all the time. Here are just a few bots that are learning how to beat us at our own games.

• French researchers are working on software that can learn how to play a game just by watching humans play -- and their software has mastered games like tic-tac-toe and connect4. More complex games like chess are a bit too difficult, but maybe someday learning algorithms will be able to take on arbitrary games. [url]
• Playing Rock-Paper-Scissors with a robot hand doesn't sound like a particularly winnable game -- especially when the robot cheats all the time. This robot only takes about a millisecond to detect what a human hand will play, so the only winning strategy for humans is to try to come up with deceptive ways to play RPS. [url]
• A panel of judges couldn't tell the difference between a computer player and a human player in an Unreal tournament (in a strange kind of Turing test). Non-player characters controlled by algorithms are getting more and more human-like, and pretty soon humans won't even need to play video games anymore.... [url]

If you'd like to read more awesome and interesting stuff, check out this unrelated (but not entirely random!) Techdirt post.

Filed Under: ai, algorithms, game algorithms, games, npc, rps, turing test, unreal